Multimodal Artificial Intelligence for Precision Oil Reduction and Sensory Reconstruction in Traditional Chinese Foods: Mechanisms, Strategies, and Future Perspectives

ABSTRACT

Traditional Chinese foods are often characterized by high lipid contents, which create a persistent tension between preserving sensory authenticity and advancing public health goals. Direct oil‐reduction strategies frequently compromise product quality because lipids serve multiple indispensable functions beyond caloric contribution, acting as heat‐transfer media, structural modulators, and carriers of fat‐soluble flavor compounds. This review highlights recent advances in multimodal artificial intelligence (AI)‐assisted oil reduction for traditional Chinese foods within an evidence‐bounded “Perception–Decision–Execution” framework. It synthesizes progress across three tightly connected domains: AI‐assisted screening and design of lipid replacers at the molecular and mesoscale levels; AI‐assisted regulation of oil‐related processing in frying, stewing, and seasoning systems; and sensory reconstruction through multimodal instrumental signals correlated with human perception. Available evidence suggests that multimodal AI may support a gradual shift from conventional trial‐and‐error oil‐reduction practices toward more data‐informed optimization by integrating real‐time sensing, predictive modeling, and adaptive process adjustment across formulation, processing, and sensory delivery. These approaches show potential for assisting the rational design of fat replacers, dynamic regulation of oil uptake, stabilization of reduced‐fat emulsified systems, and targeted compensation for losses in aroma, texture, and mouthfeel, although their effectiveness remains highly dependent on food‐matrix‐specific validation. Despite this promise, broader translation into industrial practice remains constrained by heterogeneous datasets, limited model interpretability, the lack of standardized evaluation frameworks, and persistent scale‐up challenges. Overall, AI‐assisted oil reduction may provide a useful strategy for the health‐oriented modernization of traditional Chinese foods by supporting the functional analysis, partial decoupling, and targeted reconstruction of lipid roles while preserving their cultural identity and distinctive sensory characteristics.